Intra-Cluster Distance Minimization in DNA Methylation Analysis Using an Advanced Tabu-Based Iterative k-Medoids...

Area of Science:

• Epigenetics
• Computational Biology
• Bioinformatics

Background:

• DNA methylation profiling is crucial for understanding disease mechanisms like cancer.
• Conventional clustering methods (hierarchical, k-means) are fast but often yield suboptimal results in high-throughput analyses.
• There is a need for improved clustering methodologies that maintain speed while enhancing accuracy.

Purpose of the Study:

• To introduce three novel algorithms for comprehensive high-throughput DNA methylation analysis.
• To address challenges of high-dimensionality and suboptimal clustering in methylation data.
• To improve the accuracy and reliability of epigenetic pattern identification.

Main Methods:

• A variance-based dimension reduction algorithm for high-dimensional data.
• An outlier detection algorithm to identify and manage aberrant data points.
• An advanced Tabu-based iterative k-medoids clustering algorithm (T-CLUST) to mitigate initial solution dependency.

Main Results:

• The proposed algorithms were validated on nine real DNA methylation datasets.
• T-CLUST demonstrated higher accuracy in cluster identification compared to hierarchical and k-means clustering.
• The integrated approach effectively handles high-dimensionality and outliers in methylation data.

Conclusions:

• The developed algorithms provide a robust and accurate framework for high-throughput DNA methylation analysis.
• T-CLUST offers a significant advancement over conventional clustering techniques for epigenetic studies.
• These tools can enhance the understanding of epigenetic mechanisms in diseases such as cancer.